
import pandas as pd
import glob, folium, branca, json
import numpy as np
import matplotlib.pyplot as plt
from bokeh.plotting import figure, show
from bokeh.io import output_notebook
from bokeh.models import DatetimeTickFormatter, Legend
from bokeh.layouts import column


print(np.datetime64('now'))

2021-02-16T13:19:21


# PUT EVERY COUNTRY REPORTS IN ONE DATAFRAME
files = glob.glob('data/install*country.csv')
files.sort()
ds=pd.DataFrame()
for f in files:
    ds = pd.concat([ds,pd.read_csv(f,encoding = 'utf-16')])
ds = ds.reset_index()    
ds = ds.drop(columns=['index','Daily Device Upgrades','Total User Installs','Active Device Installs','Install events','Update events','Uninstall events'])
ds['Date']=pd.DatetimeIndex(ds['Date'])
ds['Cumul Users']=ds['Daily User Installs'].cumsum()
ds['Cumul Devices']=ds['Daily Device Installs'].cumsum()
ds.tail()


for country in np.unique(ds['Country'][~ds['Country'].isna()]):
    ds[country+'-d']=ds[ds['Country']==country]['Daily Device Installs'].cumsum()
    ds[country+'-u']=ds[ds['Country']==country]['Daily User Installs'].cumsum()
ds = ds.fillna(method='ffill')
ds = ds.fillna(0)
ds.tail()


output_notebook()

countries = np.unique(ds['Country'].fillna('undef'))
labels = ["%s" % c for c in countries] 
code = 'FR'

plot_options = dict(width=1200, plot_height=400, tools='pan,wheel_zoom,box_zoom', active_scroll="wheel_zoom", x_axis_type='datetime', toolbar_location="above")
p = figure(**plot_options)

p.line(x="Date", y='Cumul Users', source=ds,legend_label='Total Users',line_color='black',line_width=3)
p.line(x="Date", y='Cumul Devices', source=ds,legend_label='Total Devices',line_color='grey',line_width=3)

p.line(x="Date", y=code+'-u', source=ds[ds['Country']==code],legend_label=code+' Users',line_color='#ff6f69',line_width=3)
p.line(x="Date", y=code+'-d', source=ds[ds['Country']==code],legend_label=code+' Devices',line_color='#88d8b0',line_width=3)
p.xaxis.formatter=DatetimeTickFormatter(hours=["%d %B %Y"],days=["%d %B %Y"],months=["%d %B %Y"],years=["%d %B %Y"])
p.legend.location = 'top_left'

show(p)


df = pd.DataFrame(ds.groupby('Country').sum()['Daily User Installs'])
color_scale = branca.colormap.linear.viridis.scale(0,20)
map_dict = df.to_dict()

def get_color(feature):
    value = map_dict['Daily User Installs'].get(feature['properties']['ISO_A2'])
    if value is None:
        return 'white' # MISSING -> white
    else:
        #print(feature['properties']['ADMIN']+' : '+str(value))
        return color_scale(value)

m = folium.Map(
    location = [0, 0], 
    tiles="cartodbpositron",
    zoom_start = 2
)

folium.GeoJson(
    data = 'countries.json',
    style_function = lambda feature: {
        'fillColor': get_color(feature),
        'fillOpacity': 0.7,
        'color' : 'None',
        'weight' : 1,
    }    
).add_to(m)
m.add_child(color_scale)
m

	Date	Package Name	Country	Daily Device Installs	Daily User Installs	Daily User Uninstalls	Cumul Users	Cumul Devices
998	2021-02-13	com.kb.android.argo	NO	0	0	0	76	81
999	2021-02-13	com.kb.android.argo	RU	0	0	0	76	81
1000	2021-02-13	com.kb.android.argo	SI	0	0	0	76	81
1001	2021-02-13	com.kb.android.argo	UA	1	1	2	77	82
1002	2021-02-13	com.kb.android.argo	US	0	0	0	77	82

	Date	Package Name	Country	Daily Device Installs	Daily User Installs	Daily User Uninstalls	Cumul Users	Cumul Devices	AU-d	...	RU-d	RU-u	SI-d	SI-u	SV-d	SV-u	UA-d	UA-u	US-d	US-u
998	2021-02-13	com.kb.android.argo	NO	0	0	0	76	81	3.0	...	5.0	5.0	1.0	1.0	1.0	1.0	0.0	0.0	7.0	5.0
999	2021-02-13	com.kb.android.argo	RU	0	0	0	76	81	3.0	...	5.0	5.0	1.0	1.0	1.0	1.0	0.0	0.0	7.0	5.0
1000	2021-02-13	com.kb.android.argo	SI	0	0	0	76	81	3.0	...	5.0	5.0	1.0	1.0	1.0	1.0	0.0	0.0	7.0	5.0
1001	2021-02-13	com.kb.android.argo	UA	1	1	2	77	82	3.0	...	5.0	5.0	1.0	1.0	1.0	1.0	1.0	1.0	7.0	5.0
1002	2021-02-13	com.kb.android.argo	US	0	0	0	77	82	3.0	...	5.0	5.0	1.0	1.0	1.0	1.0	1.0	1.0	7.0	5.0

Argo Floats android application¶

Analytics stuff¶

Installations cumulation¶

World choropleth from install¶